[00:00:01] Speaker 05:
The next argued case this morning is number 21, 1747.

[00:00:04] Speaker 05:
Mid-continent steel and wire against the United States.

[00:00:10] Speaker 05:
Mr. Marshack, when you're ready.

[00:00:37] Speaker 00:
Thank you, Your Honor.

[00:00:40] Speaker 00:
We were here three years ago before this court on the issue that we're talking about today, and a lot has happened in the past three years, and I'm very happy to be back in person next time, hopefully without a mask at all.

[00:00:54] Speaker 00:
But a lot's happened in this case also.

[00:00:56] Speaker 00:
We've had a second Federal Circuit case, the Stupp case, which really talks directly on point on the issue before us, and we believe that we've put forth before the Department of Commerce

[00:01:07] Speaker 00:
a very robust record, a much better record we had the first time we were here as to why weighted averaging is the way that you should make the determination as to whether prices are significantly different.

[00:01:22] Speaker 00:
Here, the statutory purpose is for the Commerce Department to determine whether there's a significant difference between prices in a test group and in a comparison group.

[00:01:33] Speaker 00:
To make that determination,

[00:01:35] Speaker 00:
goes through that statutory purpose, commerce uses a test which is based on the so-called Cohen's D test.

[00:01:44] Speaker 00:
It's an objective mathematical formula.

[00:01:48] Speaker 00:
And for the objective mathematical formula, the first thing that commerce does

[00:01:52] Speaker 00:
is they look at what is the absolute difference between weighted average prices in the test group and the comparison group.

[00:02:00] Speaker 00:
They calculate the weighted average price for a test group, and they calculate the weighted average price for a comparison group, and they find out the difference.

[00:02:09] Speaker 00:
So that's an absolute difference in weighted average prices.

[00:02:13] Speaker 00:
But the fact whether it's 9 cents or 25 cents, that doesn't make it significant.

[00:02:18] Speaker 00:
To make it significant, you have to compare that absolute price against what's normal in the condom in both groups.

[00:02:29] Speaker 02:
What's normal in the dispersion?

[00:02:32] Speaker 00:
Yes.

[00:02:32] Speaker 00:
What's normal.

[00:02:33] Speaker 02:
The whole point of the denominator is to give what you called and the commerce called a yardstick to decide whether the absolute difference between the two means is well within the dispersion of something.

[00:02:46] Speaker 02:
And there's an argument about something.

[00:02:48] Speaker 02:
Um, and they say just do the simple average.

[00:02:51] Speaker 02:
There are at least two alternatives to a simple average.

[00:02:54] Speaker 02:
One is the one that you're arguing about entirely, which is weighted average, which is done everywhere else and which Cohen says you are supposed to use when the two groups are different in number.

[00:03:07] Speaker 02:
Correct.

[00:03:08] Speaker 02:
What's wrong with using the standard deviation of the total population since you actually do have all of the data, right?

[00:03:17] Speaker 00:
Standard deviation.

[00:03:18] Speaker 02:
Of all the prices in the United States,

[00:03:21] Speaker 02:
uh... during the uh... period of review with this quote cohen and ellis and co talk about using uh... mixes of the two groups as a way of trying to get close to information you may not have which is standard deviation of the entire population you actually do have we have an entire population so why not use that we will be using it doesn't Cohen say you should use that when we use the weighted average we are using

[00:03:49] Speaker 00:
And that's really, in my mind, I'm trying to think mathematically here, it's really the same thing.

[00:03:56] Speaker 00:
We're using the standard deviation, when you weight the standard deviation of one group and weight the standard deviation of the other group, you're taking the standard deviation of the entire population, of every single kilogram.

[00:04:11] Speaker 00:
So what you want to do is to get the standard deviation.

[00:04:15] Speaker 02:
There hasn't been a specific argument about that here.

[00:04:19] Speaker 00:
No, no, there hasn't been, because I think what we're doing is accomplishing that same purpose.

[00:04:25] Speaker 00:
When we write the standard deviation of A and the standard deviation of B,

[00:04:31] Speaker 02:
we're literally looking at the test group and the comparison group, but once we do the weigh-in, the comparison group here is everything else.

[00:04:39] Speaker 02:
It's not two samples.

[00:04:41] Speaker 00:
No, no, no.

[00:04:43] Speaker 00:
This is by condom.

[00:04:45] Speaker 00:
So we have a group.

[00:04:48] Speaker 00:
And we have a group, let's say we have a group of 15 sales with 7,000, 7,000 kilograms.

[00:04:56] Speaker 00:
And two sales have 800 kilograms.

[00:05:00] Speaker 00:
They're very close together.

[00:05:01] Speaker 00:
And this is the example that we used.

[00:05:03] Speaker 00:
And there are 13 sales with 5,000 kilograms total.

[00:05:09] Speaker 00:
And they are more dispersed.

[00:05:12] Speaker 00:
So what we're doing is we're finding out what the standard deviation is of the 800 plus the 5,000, the 5,800.

[00:05:20] Speaker 00:
So we're finding the standard.

[00:05:22] Speaker 00:
We're pooling.

[00:05:23] Speaker 00:
And when we wait to pooling, it's the standard deviation of everything in that column, every single sale, every single kilogram.

[00:05:30] Speaker 00:
And what we're saying is to do this correctly to achieve the statutory purpose of finding whether the

[00:05:39] Speaker 00:
one group differs significantly from the other group, what you have to do is you have to, each kilogram has to have an equal weight throughout.

[00:05:50] Speaker 00:
And you have to look at the, when you look at the yardstick, you have to look at the yardstick by weighting one group and the other group, because if you use a simple average, which Congress is doing, you're taking, you know,

[00:06:04] Speaker 00:
five kilograms or ten kilograms and you're making them so much more important.

[00:06:09] Speaker 00:
You're giving them undue weight.

[00:06:10] Speaker 00:
And that's what Cohen and that's what Ellis and that's what Coe and that's what all the statistical authority and the mathematical authority say you can't do.

[00:06:20] Speaker 00:
What's wrong with a simple average when you have a situation when you have a small number or you have

[00:06:28] Speaker 00:
And I believe what Stop's talking about, when the data groups compared are small or not normally distributed and have disparate variances, then the simple average just doesn't work.

[00:06:40] Speaker 00:
And we believe that we've shown that it just doesn't work

[00:06:44] Speaker 00:
by the examples that we've used.

[00:06:46] Speaker 00:
What we did is, again, we were here before we gave the hypothetical where we had two sales, very small kilograms, very close together, and we showed when you looked at it, when you looked at it, the

[00:07:02] Speaker 00:
As you said, you were here before.

[00:07:04] Speaker 00:
Yes, Your Honor.

[00:07:05] Speaker 04:
And the panel of this court sent it back.

[00:07:08] Speaker 00:
Correct, Your Honor.

[00:07:10] Speaker 04:
And the Commerce Department went over again.

[00:07:13] Speaker 04:
Yes, Your Honor.

[00:07:14] Speaker 04:
Specialized agency and said, well, we really think it's a simple average.

[00:07:21] Speaker 04:
And the Specialized Trade Court said the same thing.

[00:07:27] Speaker 04:
How much does this need to be litigated?

[00:07:30] Speaker 00:
Well, Your Honor, I guess very fortunately for us, it was litigated in another case before the circuit, after the initial decision in this case.

[00:07:39] Speaker 00:
And that was the Stupp case, which came out, you know, less than a year ago.

[00:07:43] Speaker 00:
And Stupp, we believe, you know, litigated this matter and couldn't have been clearer by saying, we agree that there are significant concerns relating to commerce application, to Cohen's detest,

[00:07:56] Speaker 00:
when the groups are small and not normally distributed on different variances and then they said it's inaccurate when it's not normally distributed or lacking equal variances and then the court most significantly said

[00:08:11] Speaker 00:
That is because commerce uses the simple average pool standard deviation instead of the weighted average pool standard deviation.

[00:08:19] Speaker 00:
So in that case, we believe the court recognized that when you use weighted averaging, you achieve the statutory purpose.

[00:08:27] Speaker 00:
And the key is the statutory purpose is your significant difference.

[00:08:32] Speaker 00:
So the test

[00:08:34] Speaker 00:
is limited.

[00:08:36] Speaker 00:
Commerce can't do anything.

[00:08:37] Speaker 00:
They have to look at the statutory purpose to see a significant difference.

[00:08:41] Speaker 00:
And when you look at the literature, the literature says, to achieve this statutory purpose, you have to use weighted averaging to see if there's a significant difference.

[00:08:52] Speaker 00:
Because that's what the literature, that's what Professor Cohen and Cohen and Ellis talk about.

[00:09:00] Speaker 00:
Is there a significant difference?

[00:09:01] Speaker 00:
And they say, in this type of situation,

[00:09:04] Speaker 00:
And Cohen says where you have a, when you know what the population is.

[00:09:09] Speaker 02:
Can I just ask, did commerce cite other literature for its simple average?

[00:09:15] Speaker 00:
I don't believe they did.

[00:09:17] Speaker 00:
I don't remember seeing any.

[00:09:18] Speaker 00:
I don't remember anything.

[00:09:19] Speaker 00:
And I think, you know, last time your writer talked about co and, but since then, when we've looked at the literature and we cite this,

[00:09:26] Speaker 00:
In our brief, we have two direct citations where we're Cohen, and then we have Ellis also saying, look, Ellis is saying, and this is page 687, choosing among these equations requires an examination of the standard deviation of each group.

[00:09:47] Speaker 00:
If they are roughly the same, then it is reasonable to assume they are estimating a common population standard deviation, so they have to be roughly the same.

[00:09:56] Speaker 00:
But then another approach, which is recommended if the groups are dissimilar in size, is to weigh each group's standard deviation by its sample size.

[00:10:04] Speaker 00:
Cohen also talks about you can't use a simple average, and this is in the appendix to our brief, when it clearly cannot be referred to the natural population with its variance group frequencies.

[00:10:18] Speaker 00:
Also, Cohen talks about, and this is the appendix 6.735, when, however, the populations are concrete, an unequal collection of cases, the inequality should figure in the assessment of the degree of relationship.

[00:10:34] Speaker 00:
We've buttressed the argument.

[00:10:36] Speaker 00:
We think our argument is just so much, it was good last time, but it's so much better.

[00:10:41] Speaker 00:
And one of the major reasons why I believe it's so much better, aside from stuff, is we have the examples.

[00:10:46] Speaker 00:
We have gone into the data based on the facts of this case, and we've used the hypothetical we used last time.

[00:10:53] Speaker 00:
But then in our brief, and you know,

[00:10:56] Speaker 00:
This was on the record, our brief of page 23, our brief of page 25, our brief of page 26, our brief of page 27 and 28, we have four examples from the data.

[00:11:08] Speaker 00:
And when you look at the examples, and we have the dot graphs,

[00:11:11] Speaker 00:
We show that in the first two examples, the dot graphs, you have a small group where it's close together, where you have a small differential.

[00:11:23] Speaker 00:
And because that small group has a small differential, when you used

[00:11:28] Speaker 00:
a simple average that makes that small group much more important, and commerce says there's a pass.

[00:11:36] Speaker 00:
But when you use the weighted average, it puts that group in the, it gives it the appropriate weight.

[00:11:43] Speaker 00:
And the appropriate weight is based on the kilograms in that group.

[00:11:46] Speaker 00:
And when it gives it the appropriate weight, then you could visually see that there's no significant difference.

[00:11:52] Speaker 00:
So you look at the dots, and there is, I mean, anybody

[00:11:57] Speaker 00:
It's easy for me to say, anybody looking at these dot graphs would have to say that the two sales in our primary example, our brief at 23, the prices are not significantly different than the other sales.

[00:12:12] Speaker 00:
Or our brief at 25, the prices in these two sales in the test group are not significantly different because they fall right within the range of the normal disbursement of the sales.

[00:12:25] Speaker 00:
So again,

[00:12:27] Speaker 00:
you have an actual price with a difference

[00:12:31] Speaker 00:
let's say is nine cents, but you have, when you look at the disbursement of sales, the disbursement within the group, the test group, they fall right within.

[00:12:42] Speaker 00:
You could just take those two sales in the test group, put them in the comparison group, and there's no difference because they're right, they're the same.

[00:12:50] Speaker 00:
And that's what the formula is supposed to be getting at.

[00:12:53] Speaker 00:
Is it supposed to be, is there a significant difference?

[00:12:56] Speaker 00:
And it gets to the statutory purpose.

[00:12:58] Speaker 00:
And by these examples,

[00:13:01] Speaker 00:
We don't see how you get around them.

[00:13:03] Speaker 00:
I mean it just and you know, we were waiting We gave the examples in our brief we were waiting for the government or petitioner to so somehow our examples are wrong Just something's wrong with our data something's wrong with our numbers or to provide examples themselves They didn't do it.

[00:13:20] Speaker 00:
Um, and when you look at it You know, this is the application of the weighted average of

[00:13:26] Speaker 02:
Can I just add, as I understand it, commerce on the remand didn't actually find any kind of statistical literature support for what it was doing.

[00:13:38] Speaker 02:
Instead it really comes down to something seemingly very simple that, and I'm not sure I'm going to get the language right, but something like each of these sales is a rational sale for the

[00:13:55] Speaker 02:
for the foreign producer exporter to make.

[00:13:59] Speaker 02:
And each sale, therefore, or each group has to be considered as equal in stature.

[00:14:06] Speaker 02:
I don't know quite what the word was, stature, dignity, rationality, something.

[00:14:11] Speaker 02:
And therefore, group to group is what we're looking for, regardless of anything else about the numbers in the group

[00:14:23] Speaker 02:
or any basis in the statistical literature.

[00:14:26] Speaker 02:
Is that?

[00:14:27] Speaker 00:
I believe that that's what commerce says.

[00:14:29] Speaker 02:
Right.

[00:14:30] Speaker 02:
But if that theory is right, it makes all of the discussion of the statistics or what you're getting from the picture just irrelevant, which is, I think, why they don't discuss it.

[00:14:41] Speaker 00:
I think that's what they would like to say, your honor, that the method.

[00:14:45] Speaker 00:
But what they're saying is the method.

[00:14:46] Speaker 00:
What's wrong, then?

[00:14:47] Speaker 02:
What's that wrong with the theory?

[00:14:50] Speaker 00:
They're saying the methodology is key and not

[00:14:53] Speaker 00:
the actual results.

[00:14:54] Speaker 00:
And we're saying, OK, that's your methodology.

[00:14:57] Speaker 00:
How does that methodology conform to the statutory purpose to determine whether there's a significant difference in prices in those two groups?

[00:15:06] Speaker 00:
And they can't do that because once you look at the data, when you do a simple average,

[00:15:12] Speaker 00:
There's no significant difference between the prices.

[00:15:16] Speaker 00:
Just saying it doesn't make it true.

[00:15:19] Speaker 00:
You have a methodology.

[00:15:20] Speaker 00:
Well, what are the results of your methodology?

[00:15:23] Speaker 00:
I could have a theory.

[00:15:24] Speaker 00:
But if the results of my theory don't give me the correct answer, my theory goes up in smoke.

[00:15:30] Speaker 00:
How do you test the theory?

[00:15:32] Speaker 00:
You test the theory by examples.

[00:15:34] Speaker 00:
And what we've done, we've taken the data in the record

[00:15:38] Speaker 00:
And we've shown by, for example, we basically gave commerce and we've given your court information to do it for every single sale in the entire database.

[00:15:50] Speaker 00:
But we've said, okay, let's test that theory and let's look at the results.

[00:15:54] Speaker 00:
When you look at the results,

[00:15:56] Speaker 00:
Their theory falls because their theory doesn't give you something, doesn't give you an answer that shows a significant difference.

[00:16:04] Speaker 00:
Okay, they're right.

[00:16:05] Speaker 00:
But let's test it.

[00:16:07] Speaker 00:
Let's see what the results are.

[00:16:08] Speaker 00:
The results are our results.

[00:16:10] Speaker 00:
And in our results, there's no significant difference.

[00:16:13] Speaker 00:
And that's how you check a theory.

[00:16:15] Speaker 00:
So we believe their theory is wrong.

[00:16:17] Speaker 00:
They're just saying it.

[00:16:19] Speaker 00:
There's nothing to back it up except they're saying it.

[00:16:22] Speaker 00:
And when you look at the results of the theory, it's just not right as a matter of fact.

[00:16:28] Speaker 00:
So I think, again, you have to look at the literature, you have to look at the facts, and the theory has no support in the literature, it has no support in the facts.

[00:16:40] Speaker 00:
And it's a theory, great, it's a theory.

[00:16:44] Speaker 00:
I believe I'll...

[00:16:45] Speaker 00:
I think the red means I have to, this is my rebuttal time.

[00:16:49] Speaker 05:
Yes.

[00:16:50] Speaker 05:
No, we'll save your rebuttal time.

[00:16:52] Speaker 00:
Okay.

[00:16:52] Speaker 00:
Thank you, your honor.

[00:16:53] Speaker 05:
And let's hear from the other side.

[00:17:02] Speaker 05:
Ms.

[00:17:02] Speaker 05:
Cottage.

[00:17:26] Speaker 05:
Yes, when you're ready, proceed.

[00:17:29] Speaker 03:
Thank you, Your Honor.

[00:17:31] Speaker 03:
Good morning, and may it please the Court.

[00:17:34] Speaker 03:
We respectfully request that Commerce's determination to use a simple average when determining the pooled standard deviation, a part of the Cohen's D test in Commerce's differential pricing analysis, be sustained by this Court as reasonable.

[00:17:51] Speaker 03:
just as it was sustained by the trail.

[00:17:53] Speaker 05:
What's your answer to the last point that council made?

[00:18:01] Speaker 03:
Council made a number of points in the last few seconds.

[00:18:04] Speaker 03:
If I can go directly to the point about not having any support in the literature, they're wrong.

[00:18:13] Speaker 03:
The literature actually does support commerce's determination here.

[00:18:19] Speaker 03:
The whole purpose of commerce's determination in the differential pricing analysis and the Cohen's D test is to uncover masked or targeted dumping.

[00:18:33] Speaker 03:
In this case, that purpose is served by the use of a methodology that is consistent with legislative intent and directive.

[00:18:44] Speaker 03:
Congress chose to allow commerce

[00:18:48] Speaker 03:
to make a difficult determination as to whether or not there is a difference in pricing by region, purchaser, or time period.

[00:18:59] Speaker 02:
You started by saying you were, I thought you said you were going to immediately refer to literature, right?

[00:19:06] Speaker 02:
Congress said you can make a determination, but it has to be reasonable.

[00:19:13] Speaker 02:
So the reasonableness turns on whether you have

[00:19:18] Speaker 02:
valid, not just legislative intent, but valid literature.

[00:19:22] Speaker 02:
And you've relied on statistical methodologies as the foundation.

[00:19:28] Speaker 02:
And I guess the problem I'm having is I don't see that there is any support in the statistical literature for you're doing what you're doing, which means you may have to just say statistical literature, nevermind.

[00:19:41] Speaker 02:
And then say, we're going to do this for the following reason, but it has none of the backing of statistical literature.

[00:19:47] Speaker 02:
So is there backing in the statistical literature, and where exactly?

[00:19:51] Speaker 03:
Yes, Your Honor.

[00:19:52] Speaker 03:
The answer to your question is yes.

[00:19:54] Speaker 03:
The literature supports commerce's use of a simple average.

[00:19:59] Speaker 03:
For sure, the literature also supports the use of a weighted average.

[00:20:05] Speaker 02:
So where's the first part?

[00:20:06] Speaker 03:
The first part is Dr. Ellis identified three measures of effect size that could be used.

[00:20:15] Speaker 02:
I'm sorry, and just to be clear, what we're looking for is statistical literature that says when you have test groups of different sizes and you're using, and we're just talking about the denominator as a yardstick, is there literature that says when the two test groups are of different sizes, you can use the simple average?

[00:20:40] Speaker 03:
Yes, Your Honor.

[00:20:41] Speaker 02:
Okay, where's that?

[00:20:42] Speaker 03:
Dr. Cohen himself states

[00:20:45] Speaker 03:
that when the standard deviations of the two groups are not equal, the formula requires the root mean square of the standard deviation for each group.

[00:20:57] Speaker 03:
That is the square root of the mean of the two variances.

[00:21:01] Speaker 03:
Dr. Cohen goes on to state... But what page are you looking at?

[00:21:07] Speaker 02:
We have, you know, like 50 pages of Cohen in the appendix.

[00:21:15] Speaker 03:
appendix page 1103 your honor.

[00:21:24] Speaker 03:
May I continue?

[00:21:26] Speaker 02:
I want to be looking at exactly what you're looking at.

[00:21:33] Speaker 02:
I'm sorry, I thought you were, that's not Cohen.

[00:21:37] Speaker 02:
We have the Cohen book.

[00:21:39] Speaker 02:
Oh, I'm sorry, your honor.

[00:21:40] Speaker 02:
We're in the Cohen book.

[00:21:47] Speaker 02:
And I think that the specific passage that this cites to is from a page in colon 44 where NA is equal to NB, which is exactly the assumption that is not true here.

[00:22:12] Speaker 03:
Dr. Cohen goes on to state, Your Honor, that the unequal variability need not affect the conception of D developed in Section 2.2.

[00:22:26] Speaker 02:
Just to be clear, so the thing, this is part of, this is the commerce decision, right?

[00:22:33] Speaker 02:
This 1103, right?

[00:22:35] Speaker 02:
And it cites page 43, 44.

[00:22:38] Speaker 02:
That's exactly a page, appendix 745, which is Cohen.

[00:22:44] Speaker 02:
And this is case number two.

[00:22:48] Speaker 02:
The two standard deviations are different.

[00:22:52] Speaker 02:
And NA equals NB.

[00:22:55] Speaker 02:
Test group size, same as comparison group size, that's not what we have.

[00:22:58] Speaker 02:
Of course if they're the same size you can use a simple average, but that's not what we have.

[00:23:03] Speaker 02:
So where does Cohen or any other literature say you can use a simple average to get at this when the two groups are different in size?

[00:23:23] Speaker 02:
I understand that Commerce has an argument that basically says, statistics be damned.

[00:23:30] Speaker 02:
We just have a theory about equal stature of the two groups.

[00:23:33] Speaker 02:
And it doesn't matter if we're doing anything with an accepted statistical foundation.

[00:23:39] Speaker 02:
But you've also suggested that there is an acceptable statistical foundation for this.

[00:23:45] Speaker 02:
And that's what I am simply failing to see.

[00:23:49] Speaker 03:
OK, Your Honor.

[00:23:55] Speaker 03:
Dr. Cohen did state that the unequal variability need not affect the conception of D developed in section 2.2.

[00:24:04] Speaker 03:
Given that there is a difference between the variance of one test group and the variance of the other test group, we merely are using a kind of average within population standard deviation to standardize the differences between the means.

[00:24:30] Speaker 03:
Commerce made a determination that the literature of record, Your Honor, supported the use of a simple average.

[00:24:39] Speaker 03:
They fully recognize and appreciate the fact that a weighted average is also recognized in that literature.

[00:24:47] Speaker 03:
But for purposes of the determination that commerce is making, commerce determined that a simple average is more appropriate for precisely the reason that you stated earlier when

[00:24:59] Speaker 03:
PT's counsel was arguing, and that is because the statute requires commerce to look at the sales to region, time period, and purchaser to assess those sales as against all other comparable sales of the merchandise during that period.

[00:25:26] Speaker 03:
You have two separate and distinct groups.

[00:25:30] Speaker 03:
Their behavior, the pricing behavior itself is what is being looked at at this stage.

[00:25:37] Speaker 03:
The weighted averaging is in the numerator, but the denominator deals with specifically the pricing behavior.

[00:25:47] Speaker 03:
This is not sampling.

[00:25:49] Speaker 03:
This is all of the data.

[00:25:52] Speaker 03:
This is not a presumption.

[00:25:54] Speaker 03:
This is an assessment of pricing behavior to determine whether or not there are pricing differences.

[00:26:01] Speaker 03:
And for that purpose, commerce deems the two groups to be equally relevant.

[00:26:08] Speaker 03:
and to not do weighted averaging because it would steer the results towards whichever group, whether that be the test group or the comparison group, had the most transactions or observations.

[00:26:23] Speaker 03:
And that's not the purpose.

[00:26:25] Speaker 03:
The purpose is to assess the pricing behavior to this purchaser versus the pricing behavior to all other purchasers.

[00:26:34] Speaker 03:
the pricing behavior to this region versus all other regions, and the pricing behavior at this specific time versus all other times.

[00:26:45] Speaker 03:
And Commerce has determined that a simple average is the appropriate means of doing that.

[00:26:50] Speaker 03:
The trial court as well, which is highly specialized in this area of practice as well, determined that Commerce's determination was reasonable.

[00:27:03] Speaker 03:
It this, this for me is rocket science because I am not analytically inclined.

[00:27:10] Speaker 03:
However, commerce has been charged by Congress.

[00:27:14] Speaker 02:
Why didn't it provide an explanation then that actually identified something in the statistics that supports this as opposed to, I mean, I frankly don't really understand what relation to the very pinpoint statistical purpose of the denominator.

[00:27:32] Speaker 02:
this business about equal stature or something of the groups has.

[00:27:38] Speaker 02:
The numerator says you've got two groups, you've got your means.

[00:27:42] Speaker 02:
The means are different.

[00:27:44] Speaker 02:
It's 9 cents, it's 25 cents, whatever it is, and now you're trying to figure out is that a significant

[00:27:50] Speaker 02:
difference.

[00:27:51] Speaker 02:
And the whole point of Cohen is to say the only way you can figure out if the difference between the two groups is different is to look at what the dispersion within the whole population is or within the dispersion of the two groups.

[00:28:08] Speaker 02:
And Cohen and Ellis and Coe

[00:28:11] Speaker 02:
uniformly say that to get the right yardstick for determining if that number in the numerator is just an ordinary piece of the ordinary dispersion is to see what the dispersions are in the pool.

[00:28:30] Speaker 02:
So you have to look at the statistical literature.

[00:28:32] Speaker 02:
It's not that any sale is of greater dignity than any other sale.

[00:28:36] Speaker 02:
I'm not even sure what that means as a statistical matter.

[00:28:42] Speaker 03:
My time is up, Your Honor, but may I respond?

[00:28:46] Speaker 03:
Commerce did say that any sale is more relevant than any other sale.

[00:28:51] Speaker 03:
The statute requires that Commerce divide the sales up based on purchaser region and time period.

[00:29:01] Speaker 03:
And Commerce did that.

[00:29:03] Speaker 03:
And how do you analyze the sales

[00:29:06] Speaker 03:
from one purchaser as against all other purchasers when the purpose of this is that designed to unmask targeted dumping.

[00:29:14] Speaker 03:
And with respect to that, your honor, each and every one of the expert literature sources that are on the record state that context matters.

[00:29:24] Speaker 03:
The purpose of your investigation matters.

[00:29:28] Speaker 03:
And each one of them chooses a denominator that meets the goals for the purposes that they are conducting their study.

[00:29:37] Speaker 03:
And commerce, what commerce did here is no different.

[00:29:41] Speaker 03:
It used a test for purposes of unmasking targeted dumping.

[00:29:49] Speaker 03:
And it used it within the scope of its responsibilities under the statute.

[00:29:55] Speaker 03:
And with that, Your Honor, I would request that the court sustain Congress's determination as reasonable.

[00:30:03] Speaker 03:
Nothing has been said here today to state that it is unreasonable or demonstrate that it is unreasonable.

[00:30:10] Speaker 03:
Thank you, Your Honor.

[00:30:23] Speaker 05:
Now, my list says that the next speaker is Adam Gordon.

[00:30:29] Speaker 05:
Is that correct?

[00:31:16] Speaker 05:
Okay.

[00:31:17] Speaker 05:
If you're ready, Mr. Gordon, please proceed when ready.

[00:31:24] Speaker 01:
Good morning and may it please the court.

[00:31:27] Speaker 01:
I'd just like to offer a few comments to follow up on what the government presented a few minutes ago.

[00:31:34] Speaker 01:
First, I would like to point out to the court that

[00:31:38] Speaker 01:
all the literature on the record before the court, when it talks about waiting, it always talks about waiting in terms of the number of observations.

[00:31:45] Speaker 01:
And that's even recognized in the stuff decision, as we've discussed in our brief.

[00:31:50] Speaker 01:
The PT, the Taiwan respondents, are asking the court to wait by

[00:31:56] Speaker 01:
volume that is not supported anywhere in the literature and in fact as we and as the government showed in its remand determination and as we discussed doing so which is not only not supported by the literature but waiting in the context of calculating the pooled standard deviation whether it's by volume or uh... any other metric uh... would introduce distortions into the into the calculation can i just say

[00:32:24] Speaker 02:
I guess two things I wanted you to respond to.

[00:32:28] Speaker 02:
One is this point about counting transactions is not part of the commerce decision, right?

[00:32:35] Speaker 02:
And second, I thought when we were here three years ago, one of the reasons that commerce gave in its underlying decision for

[00:32:44] Speaker 02:
Um, for, uh, what it did was to say, if you count by transactions, it's very manipulable.

[00:32:52] Speaker 02:
Cause you can take, you can make 50 individual sales of one kilogram a piece or one sale of 50 kilograms.

[00:32:59] Speaker 02:
And commerce said that would be terribly unreliable because it would be manipulable.

[00:33:03] Speaker 02:
So obviously you can't count by numbers of transactions.

[00:33:07] Speaker 02:
And yet I think you're making the argument you should now return to something commerce didn't do and had a pretty good reason not to do.

[00:33:14] Speaker 01:
Not at all, Your Honor.

[00:33:15] Speaker 01:
My point is that my initial point is that the simple average is the way to go.

[00:33:20] Speaker 01:
But if you were to weight average at all,

[00:33:25] Speaker 01:
you have to do it by the number of observations.

[00:33:28] Speaker 01:
The literature supports both using a simple average or using a weighted average in different variations.

[00:33:35] Speaker 01:
Nothing in the literature supports an approach to weighting based on volume.

[00:33:40] Speaker 01:
The reason for that is illustrated in commerce's

[00:33:43] Speaker 01:
uh... own uh... uh... own own uh... track remand termination appendix six which has a has a uh... a listing of an analysis of the top twenty five observations of control numbers by region and it shows what happens when you wait the calculation of the pooled standard deviation by volume what happens there is

[00:34:04] Speaker 01:
understanding that you have test groups and comparison groups many times but not always the test group is much smaller than the comparison group so if you wait by volume you will always be giving greater weight to the pricing into the variance in the comparison group conversely if your test group is ever larger than the comparison group you'll

[00:34:25] Speaker 01:
give greater weight to that group.

[00:34:27] Speaker 01:
So if you look in that appendix, it shows you the way the weights of the two parts of the denominator can differ depending on the volume.

[00:34:36] Speaker 01:
And they range dramatically from having the, sometimes they're equal, sometimes it's 50-50, sometimes it's

[00:34:41] Speaker 01:
I'm making these numbers up, 10, 90.

[00:34:43] Speaker 01:
Sometimes it's 90, 10.

[00:34:45] Speaker 01:
But what you would be doing if you used volume to weight in that part of the calculation is to introduce a level of distortion within that calculation that you would have no idea.

[00:34:54] Speaker 01:
It would totally negate the goal of predictability and reliability within that calculation.

[00:35:01] Speaker 02:
It's that last part that I don't really understand.

[00:35:04] Speaker 02:
Why is that distortion?

[00:35:06] Speaker 02:
And in particular, why not just

[00:35:08] Speaker 02:
Yes.

[00:35:09] Speaker 02:
Why is that distortion?

[00:35:10] Speaker 02:
Why isn't that exactly what this statistical test that commerce invokes as the basis for its decision say that you in fact are supposed to do in the denominator?

[00:35:25] Speaker 01:
Okay.

[00:35:25] Speaker 01:
So let me, let me.

[00:35:27] Speaker 01:
Put a pin in your last comment.

[00:35:29] Speaker 01:
I don't think the literature says you have to do it.

[00:35:31] Speaker 01:
The literature allows the investigation.

[00:35:34] Speaker 02:
I'll say that it's the only thing that the statistical literature affirmatively advances when the groups are different sizes.

[00:35:47] Speaker 02:
Or else, Commer surely would have cited something that said that.

[00:35:50] Speaker 01:
In fact, they did, Your Honor.

[00:35:51] Speaker 01:
I apologize.

[00:35:52] Speaker 01:
I respectfully disagree with that.

[00:35:54] Speaker 01:
If you look at page 12 of our brief,

[00:35:56] Speaker 01:
It's appendix page number 1023, citing Cohen at page 361.

[00:36:00] Speaker 01:
It is a comment on doing a, here's the difference between an experimental versus an observational investigation.

[00:36:10] Speaker 01:
In that context, they're looking at an observational investigation of the attitudes of three different religious groups toward the United Nations.

[00:36:17] Speaker 01:
Apparently, this is a well-known example.

[00:36:21] Speaker 01:
And within that population, because you're looking at the population as a whole, the investigator doesn't have control over the sample size, unlike in an experimental setting where you're sampling, which is what all these models are based on.

[00:36:35] Speaker 01:
So in that context, Cohen speaks to the goal, in this case, of looking at an abstract characteristic of the subpopulation that they've sampled.

[00:36:48] Speaker 01:
And he talks about, and I'll read this, that the n squared, the eta squared, a measure of effect size from equal sampling, would have to be interpreted as the proportion

[00:36:59] Speaker 01:
of AUN, Average United Nations Attitude to the United Nations, variance associated with artificially equiprobable religious group membership.

[00:37:07] Speaker 01:
Here's the key point.

[00:37:07] Speaker 01:
The equal sampling of eta squared is not objectionable if the investigator wishes to consider membership in a given religious group as an abstract effect quite apart from the relative frequency with which that effect, i.e.

[00:37:21] Speaker 01:
that religious group, occurs in the population.

[00:37:25] Speaker 01:
So there, and I'm going to pick up on a point from Taiwan respondents' reply brief, they accused us of taking this out of context and leaving off an important following phrase.

[00:37:36] Speaker 01:
There's a following phrase which says, but of course you can't extrapolate that to the population as a whole given the unequal or the varying subpopulations within it.

[00:37:46] Speaker 01:
That's a red herring.

[00:37:47] Speaker 01:
It's irrelevant because another key point here is this deals with sampling from a population.

[00:37:54] Speaker 01:
And commerce is dealing with the entire population.

[00:37:57] Speaker 01:
So a constraint on your ability to extrapolate a characteristic from a sample size where your sample might be artificially equal, even though within the whole population, the different religious groups aren't equally represented.

[00:38:11] Speaker 01:
In this case, people of Jewish heritage are much a smaller group than people of Protestant or Catholic heritage.

[00:38:17] Speaker 01:
You can't extrapolate necessarily from the small subsample you've artificially created to the entire population.

[00:38:24] Speaker 01:
It's irrelevant because here we're dealing with an entire population.

[00:38:26] Speaker 02:
So as government council... So what makes that example the right one to draw on for the current problem?

[00:38:38] Speaker 01:
Well, it's an example of Cohen speaking to the validity of using a simple average.

[00:38:43] Speaker 02:
in particular circumstances.

[00:38:44] Speaker 02:
So why are those particular circumstances ones that match the circumstances of the inquiry here?

[00:38:50] Speaker 01:
Because we are looking at an abstract characteristic of the population.

[00:38:54] Speaker 02:
What does the word abstract mean in that context?

[00:38:57] Speaker 01:
So it means it is a characteristic in this case pricing behavior

[00:39:02] Speaker 01:
essentially one step removed from the actual pricing and the volumes and the sales and the customers.

[00:39:10] Speaker 01:
Commerce is charged with identifying patterns of prices that differ significantly by region customer time period.

[00:39:17] Speaker 01:
Right.

[00:39:17] Speaker 02:
So it's going to look at first quarter prices and say the average in the first quarter differs by, I don't know, five cents a kilogram from the average in Q2 through four.

[00:39:31] Speaker 02:
And then the question is, is five cents, does that differ significantly?

[00:39:35] Speaker 02:
That's the statutory term.

[00:39:37] Speaker 02:
And the question and the answer that Cohen's D tells you to look at is, well, what does the dispersion within the, you know, the thousand sales in Q1, the thousand sales in Q2, et cetera, look like?

[00:39:52] Speaker 02:
And in order to do that, you need to figure out how are you going to assess the dispersion, right?

[00:40:02] Speaker 01:
Yes.

[00:40:03] Speaker 02:
Right, so how is it that this, what's abstract about this?

[00:40:10] Speaker 01:
Because you're assessing, so step back for a moment, you have a numerator,

[00:40:15] Speaker 01:
that is calculated as a difference between the weighted average pricing for the test group and the weighted average pricing for the comparison.

[00:40:21] Speaker 01:
Important footnote to that.

[00:40:22] Speaker 01:
Look at that.

[00:40:24] Speaker 01:
The volume of the sales, the weight of the volume, which is what the respondents go on and on about how important that is, it's already taken into account in the calculation because your average prices of the numerator already reflect weighted average pricing.

[00:40:39] Speaker 01:
Then commerce, so you have there a difference.

[00:40:41] Speaker 01:
That difference, as I think Your Honor even pointed out, what's it mean?

[00:40:45] Speaker 01:
It is an abstract difference.

[00:40:46] Speaker 01:
It's a measure of the difference between one weighted average price and another.

[00:40:50] Speaker 01:
So there you have this data point sitting up in the numerator.

[00:40:52] Speaker 01:
How do you make sense of it?

[00:40:53] Speaker 01:
You have to make sense of it by calculating the pool standard deviation.

[00:40:57] Speaker 02:
But here- Or as Cohen says, if you actually have all the data, what's wrong with using the standard deviation of the entire year's worth of sales?

[00:41:05] Speaker 02:
Well, that may be- That tells you actually what the dispersion is across

[00:41:09] Speaker 02:
across the whole population.

[00:41:12] Speaker 01:
I have to say, Your Honor, I hadn't thought about that until you mentioned it the first time, and that's not before the court on appeal.

[00:41:18] Speaker 01:
It's an interesting idea, but the issue before the court on appeal is whether commerce's use of a simple average to calculate the pool standard deviation is reasonable.

[00:41:27] Speaker 01:
The responder is saying, no, it's not.

[00:41:29] Speaker 01:
Use our approach, which has no foundation in the literature and which would introduce distortions into the calculation,

[00:41:34] Speaker 01:
that would debase this goal of calculating on a reliable, predictable, consistent basis by using the simple average, it would distort this goal.

[00:41:46] Speaker 01:
Based on the fact that you have

[00:41:48] Speaker 01:
a test and comparison group drawn from the population as a whole that will always be different in size.

[00:41:53] Speaker 01:
Not always.

[00:41:54] Speaker 01:
No, I should say that because there are times when they're equal.

[00:41:56] Speaker 01:
But in most cases, I think it's safe to assume they're going to be different in size.

[00:42:00] Speaker 01:
And if you start weighting it on any basis, you're automatically going to overweight one side and underweight the other side.

[00:42:08] Speaker 01:
And if you look at the data that I cited to in the appendix, that shows you how different the weighting ends up being.

[00:42:17] Speaker 01:
It's not consistent.

[00:42:19] Speaker 01:
So it's not consistent.

[00:42:21] Speaker 01:
It's always going to be different based on how large one group versus another is.

[00:42:25] Speaker 01:
And as I said, you've already weighted.

[00:42:27] Speaker 01:
You've already taken into account the significance of the size of the sales with respect to pricing by calculating the weighted average prices in the numerator.

[00:42:41] Speaker 05:
Any more questions?

[00:42:44] Speaker 01:
May I just wrap up with just a couple of quick points?

[00:42:46] Speaker 01:
I think there's one analogy that would be useful to the court.

[00:42:49] Speaker 01:
So with respect to the examples that PT has given here, they offered four examples to show they say visually it's intuitive.

[00:42:57] Speaker 01:
If you look at these examples, it makes so much sense these prices aren't different.

[00:43:02] Speaker 01:
Those are four examples drawn from a database of hundreds and hundreds and hundreds of calculations.

[00:43:07] Speaker 01:
That's actually in

[00:43:10] Speaker 01:
uh... if you look at appendix page one four four one it allows you to calculate the number of comparisons that were actually by commerce i think the correct way to conceptualize this is as if you're thinking about a commerce department and i don't think it is set when they run their anti-dumping calculations you'll have hundreds thousands of of a of of individual control number calculations within any given database

[00:43:33] Speaker 01:
Regardless of the outcome, some of those calculations will show dumping, some will show not dumping, some will show dumping by extraordinary amounts, some will show no dumping by extraordinary amounts.

[00:43:44] Speaker 01:
At the end of the day, that's why you look at the data as a whole to calculate an aggregated final result.

[00:43:50] Speaker 01:
And in fact,

[00:43:52] Speaker 01:
The respondents say that you have to look at the results.

[00:43:56] Speaker 01:
Well, here, yes, the results speak for themselves.

[00:43:57] Speaker 01:
When you did the Cohen's D analysis, when you did the differential pricing analysis, it ended up showing a difference that couldn't be taken into account by using the standard methodology.

[00:44:08] Speaker 01:
And it resulted in an above-diminished dumping margin.

[00:44:12] Speaker 01:
That shows the results of the methodology as a whole.

[00:44:15] Speaker 01:
not for hand-picked samples, and that's not the way you judge these things.

[00:44:21] Speaker 01:
You have to look at the database as a whole.

[00:44:22] Speaker 01:
Thinking about it like the way you look at an anti-dumping database, I think, is a very useful analogy.

[00:44:27] Speaker 01:
You can't just get hung up on four examples.

[00:44:29] Speaker 01:
You have to think about the results of the analysis as a whole.

[00:44:32] Speaker 01:
And as government counsel said, there's nothing before the court that has shown that commerce's approach is unreasonable.

[00:44:38] Speaker 01:
And conversely, there is ample evidence on the record

[00:44:40] Speaker 01:
to show that what the respondents are suggesting is unsupported at all and will lead to distorted results.

[00:44:47] Speaker 01:
The only reason they've offered in favor of it is it gives them a non-dominant margin.

[00:44:51] Speaker 01:
It gives them a dominant margin.

[00:44:53] Speaker 01:
So with that, I would ask that the court please affirm Congress's real results.

[00:44:57] Speaker 01:
Thank you.

[00:44:57] Speaker 05:
Thank you, counsel.

[00:45:08] Speaker 05:
Mr. Marczak, when you're ready for rebuttal.

[00:45:18] Speaker 00:
There was a lot that was said in the last couple of minutes.

[00:45:23] Speaker 00:
I just want to get back again.

[00:45:25] Speaker 00:
And I think from your honor's questions, I think you understand what the case is about, and you understand the difference.

[00:45:31] Speaker 02:
Can you address the UN example, please?

[00:45:34] Speaker 00:
I'm sorry, the example.

[00:45:35] Speaker 02:
The United Nations examples, oversampling the Jewish population.

[00:45:39] Speaker 00:
Yes, the United Nations example, that was something that

[00:45:45] Speaker 00:
The petitioner put it into its brief and in our reply brief, we put an appendix, the three pages from Cohen that was talking about it.

[00:45:55] Speaker 00:
And the three pages, again, reading from Cohen.

[00:45:59] Speaker 00:
And we start with the quotation that petitioner wanted you to focus on.

[00:46:04] Speaker 00:
And that talked about when there's sampling.

[00:46:10] Speaker 00:
And when you could use equal sampling.

[00:46:12] Speaker 00:
And what Cohen said is, and I'll quote from Cohen, equal sampling is not objectionable if the investigator wishes to consider membership in a given religious group as an abstract effect quite apart from the relative frequency, which would not affect that religious group occurs in the population.

[00:46:30] Speaker 00:
So that's what was quoted by Petitioner.

[00:46:32] Speaker 00:
So there you're talking about you can use equal sampling if you're talking about an abstract effect.

[00:46:38] Speaker 00:
And an abstract effect could be you could use

[00:46:40] Speaker 00:
could be statistical sampling, for example.

[00:46:42] Speaker 00:
That could be an abstract effect.

[00:46:44] Speaker 00:
Here, one thing we're not doing is we're not talking about an abstract effect.

[00:46:48] Speaker 00:
We're talking about an actual, natural population where we have all the data in the groups.

[00:46:56] Speaker 00:
The abstract effect is when you're doing sampling.

[00:46:59] Speaker 00:
Cohen, as a social scientist, generally was talking about sampling.

[00:47:03] Speaker 00:
and abstract effects, and there's a possibility, you know, he's saying you can use equal sampling if you're talking about abstract effects when you're talking about statistical sampling.

[00:47:13] Speaker 00:
But the last clause

[00:47:15] Speaker 00:
From Cohen, which Petitioner did not cite and which now he tries to distinguish, Cohen says, but it clearly cannot be referred to the natural population with its varying group frequencies.

[00:47:27] Speaker 00:
Now, what does that mean?

[00:47:29] Speaker 00:
The phrase clearly cannot be referred to the natural population with its varying group frequencies.

[00:47:35] Speaker 00:
Cohen writes, as a mathematician, not as a lawyer, but when we're talking about a natural population, we're talking about exactly what we have here.

[00:47:45] Speaker 00:
A natural population is when we have all the data and we have all the data because we know

[00:47:55] Speaker 00:
the weight of every sale, and we know the quantity.

[00:47:59] Speaker 00:
We know the price of every sale.

[00:48:01] Speaker 00:
We have 100% of the data.

[00:48:03] Speaker 00:
And because we have the 100% of data, we're talking about a natural population rather than some sample.

[00:48:09] Speaker 00:
And we're talking about a natural population with varying group frequencies.

[00:48:14] Speaker 00:
And the other thing we have here, we definitely have varying group frequencies.

[00:48:18] Speaker 00:
We have some groups.

[00:48:20] Speaker 00:
where you have a standard deviation is very small, and some groups where a standard deviation is very large.

[00:48:26] Speaker 00:
So what Cohen is saying, and I think the only way you can construe Cohen reasonably, that is that equal sampling, you cannot use equal sampling, clearly cannot be referred to when you have a natural population with its varying group frequencies.

[00:48:43] Speaker 00:
And interpreting Cohen in that way would be basically

[00:48:48] Speaker 00:
The same way that Ellis interpreted Cohen, the same way that Coe interpreted Cohen, the same way that the Stupp Court, which also referred

[00:48:59] Speaker 00:
to some precedent, which wasn't even cited in the briefs, read the literature.

[00:49:04] Speaker 00:
So the literature, the extensive literature, and that's the term that the Stupp Court used, describes the problems associated with applying the Coen's D tests, data that are not normally distributed with or lacking equal variances.

[00:49:17] Speaker 00:
There's an incredible amount of literature out there.

[00:49:19] Speaker 00:
There's probably more literature on Coen's D than there is about anti-dumping.

[00:49:24] Speaker 00:
As lawyers, we've read a lot of it.

[00:49:27] Speaker 00:
We know the court has read a lot of it, probably even more than us.

[00:49:32] Speaker 00:
But when you look at the literature, it seems to me that the extent of literature, the preponderance, which Cohen was saying, is when you have a natural population,

[00:49:43] Speaker 00:
When you have small groups, when you have the deviation that is different between the groups, you just can't use a simple averaging.

[00:49:55] Speaker 00:
And let me just finish talking about the examples.

[00:49:58] Speaker 00:
petitioners council will say the examples don't matter because we just cherry-picked, really saying we cherry-picked examples.

[00:50:04] Speaker 00:
Well, the examples are the data.

[00:50:08] Speaker 00:
Petitioners didn't put on the examples.

[00:50:10] Speaker 00:
The government didn't put on the examples.

[00:50:11] Speaker 00:
We gave the court.

[00:50:12] Speaker 00:
We gave the petitioners.

[00:50:14] Speaker 00:
We gave data for every single data point.

[00:50:18] Speaker 00:
We put that, it's on the record, the simple average and the weighted average of every single data point in that record.

[00:50:25] Speaker 00:
And we picked these four because they were illustrative, because you could see it.

[00:50:29] Speaker 00:
But you could take any single condom and it's going to be any single condom.

[00:50:34] Speaker 00:
It's going to give you the same theoretical, the same theoretical decision.

[00:50:40] Speaker 00:
It's going to show that when you have a small group,

[00:50:43] Speaker 00:
and you have a small group with a small standard deviation, and you give that equal weight to a larger group with a larger standard deviation, you're giving unequal weight to those sales in the small group.

[00:50:58] Speaker 00:
And you're giving that small group incredible power to turn a pass into a no pass or a no pass into a pass.

[00:51:06] Speaker 00:
And those results are contrary to the statutory directive

[00:51:12] Speaker 00:
And the key, this is in the statute, that you have to show a significant difference.

[00:51:17] Speaker 00:
So commerce doesn't have unabated discretion.

[00:51:20] Speaker 00:
Commerce has to show a significant difference.

[00:51:24] Speaker 00:
And they can't just say, well, my methodology is good because I like my methodology.

[00:51:29] Speaker 00:
There's a statutory directive to show a significant difference.

[00:51:32] Speaker 00:
And we've shown that weighted averaging conforms to that statutory purpose.

[00:51:39] Speaker 00:
And simple averaging by the examples, by the literature, by the way commerce treats weighted averaging in every other area of law.

[00:51:48] Speaker 00:
Decision after decision on weighted averaging, weighted averaging is better.

[00:51:51] Speaker 00:
It's better for margins.

[00:51:53] Speaker 00:
It's better for just about everything else.

[00:51:56] Speaker 00:
And all of a sudden, commerce says, well, for this case, we don't use simple averaging because

[00:52:02] Speaker 00:
because we don't want them.

[00:52:04] Speaker 00:
That doesn't conform to the statutory purpose, Your Honor.

[00:52:07] Speaker 00:
Thank you.

[00:52:08] Speaker 05:
Gail's for counsel.

[00:52:10] Speaker 05:
Thank you.

[00:52:10] Speaker 05:
Thank you all.

[00:52:12] Speaker 05:
The case is under submission.